Qin Opera video denoising algorithm based on attention mechanism

SHI Qingaoxue; YANG Chaoran; LIU Xinda; GENG Guohua

doi:10.16152/j.cnki.xdxbzr.2025-01-014

您当前的位置：

首页 >

文章列表页 >

Qin Opera video denoising algorithm based on attention mechanism

The digital protection of cultural heritage | 更新时间：2025-02-08

- Qin Opera video denoising algorithm based on attention mechanism
- Journal of Northwest University (Natural Science Edition) Vol. 55, Issue 1, Pages: 168-179(2025)
- 作者机构：
  
  1.西北大学　文化遗产数字化国家地方联合工程研究中心，陕西　西安　 710127
  2.西北大学　可视化技术研究所，陕西　西安　 710127
- 作者简介：
- 基金信息：
- DOI：10.16152/j.cnki.xdxbzr.2025-01-014
  CLC： TP391
- Published：25 February 2025，
  
  Received：15 October 2024，
- 稿件说明：
移动端阅览
SHI QINGAOXUE, YANG CHAORAN, LIU XINDA, et al. Qin Opera video denoising algorithm based on attention mechanism. [J]. Journal of northwest university (natural science edition), 2025, 55(1): 168-179.
DOI：

SHI QINGAOXUE, YANG CHAORAN, LIU XINDA, et al. Qin Opera video denoising algorithm based on attention mechanism. [J]. Journal of northwest university (natural science edition), 2025, 55(1): 168-179. DOI： 10.16152/j.cnki.xdxbzr.2025-01-014.

摘要

秦腔，作为中国传统戏曲艺术的瑰宝，拥有深厚的历史底蕴。然而，秦腔早期的影像资料常受噪声和失真影响，导致画质不佳，严重妨碍了秦腔数字档案的保存品质。目前应用的视频去噪技术在处理秦腔那色彩丰富、纹理复杂的服饰时，往往没有充分利用视频帧序列的时间连贯性，使得去噪效果并不理想，难以有效保留视频帧的核心特征。基于注意力机制的秦腔视频去噪算法开展研究，针对现有视频去噪算法忽略帧间时序相关性导致效果不佳的问题，提出了一种新的视频去噪算法，该算法利用双门控注意力机制进行时序信息的融合。首先，通过时序融合模块，将视频连续帧的时序信息进行有效整合；其次，利用双门控注意力去噪网络精确识别并消除时序上的噪声；最后，通过多头交互注意力精炼模块进一步细化特征，以消除去噪过程中可能产生的伪影并恢复丢失的细节，从而提升去噪后图像的质量。实验结果表明，与DVDNet、ViDeNN以及FastDVDNet等现有方法相比，该方法可以更好地利用视频的时序信息，达到干净且高效的秦腔视频去噪效果。

Abstract

Qin Opera

as a treasure of Chinese traditional theatre art

has a profound historical heritage. However

the early video materials of Qin Opera are often affected by noise and distortion

resulting in poor picture quality

which seriously hampers the preservation quality of Qin Opera digital archives. Currently applied video denoising techniques often do not make full use of the temporal coherence of the video frame sequence when dealing with the colorful and complex texture of Qin Opera's costumes

which makes the denoising effect unsatisfactory and makes it difficult to effectively retain the core features of the video frames. In this paper

we carry out research on the Qin Opera video denoising algorithm based on the attention mechanism

and the main research contents are as follows: Aiming at the existing video denoising algorithms ignoring the temporal correlation between frames which leads to the problem of poor effect

we propose a new video denoising algorithm

which makes use of the double gating attention mechanism for the fusion of the temporal sequence information. The algorithm firstly integrates the timing information of consecutive video frames effectively through the timing fusion module; then accurately identifies and eliminates the timing noise using the dual-gated attention denoising network; finally

the features are further refined through the multi-head interactive attention refining module to eliminate the artifacts that may be generated during the denoising process and recover the lost details

to enhance the quality of the denoised image. The experimental results demonstrate that compared with existing methods such as DVDNet

ViDeNN

and FastDVDNet

this method can make better use of the timing information of the video to achieve clean and efficient denoising of Qin Opera videos.

关键词

秦腔视频去噪注意力机制时序融合

Keywords

Qin Operavideo denoisingattention mechanismtemporal fusion

references

周明全, 耿国华, 武仲科. 文化遗产数字化保护技术及应用[M]. 北京: 高等教育出版社, 2011.

耿国华, 何雪磊, 王美丽, 等. 文化遗产活化关键技术研究进展[J]. 中国图象图形学报, 2022, 27(6): 1988-2007.

GENG G H, HE X L, WANG M L, et al. Research progress on key technologies of cultural heritage activation[J]. Journal of Image and Graphics, 2022, 27(6): 1988-2007.

TICO M. Multi-frame image denoising and stabilization[C]//2008 16th European Signal Processing Conference. Lausanne: IEEE, 2008: 1-4.

LIU X, ZHU Y, LIU L, et al. Feature-suppressed contrast for self-supervised food pre-training[C]//Proceedings of the 31st ACM International Conference on Multimedia. Ottawa: ACM, 2023: 4359-4367.

MAGGIONI M, HUANG Y, LI C, et al. Efficient multi-stage video denoising with recurrent spatio-temporal fusion[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Virtual: IEEE, 2021: 3466-3475.

DAVY A, EHRET T, MOREL J M, et al. A non-local CNN for video denoising[C]//2019 IEEE International Conference on Image Processing (ICIP). Taipei: IEEE, 2019: 2409-2413.

TASSANO M, DELON J, VEIT T. Dvdnet: A fast network for deep video denoising[C]//2019 IEEE International Conference on Image Processing (ICIP). Taipei: IEEE, 2019: 1805-1809.

XUE T, CHEN B, WU J, et al. Video enhancement with task-oriented flow[J]. International Journal of Computer Vision, 2019, 127(8): 1106-1125.

WANG C, ZHOU S K, CHENG Z W. First image then video: A two-stage network for spatiotemporal video denoising[EB/OL]. (2020-01-22) [2024-06-20]. https://arxiv.org/abs/2001.00346v2https://arxiv.org/abs/2001.00346v2.

MEHTA S, KUMAR A, REDA F, et al. Evrnet: Efficient video restoration on edge devices[C]//Proceedings of the 29th ACM International Conference on Multimedia. Virtual: ACM, 2021: 983-992.

TASSANO M, DELON J, VEIT T. Fastdvdnet: Towards real-time deep video denoising without flow estimation[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020: 1354-1363.

VAKSMAN G, ELAD M, MILANFAR P. Patch craft: Video denoising by deep modeling and patch matching[C]//2021 IEEE/CVF International Conference on Computer Vision. Montreal: IEEE, 2021: 2157-2166.

MAGGIONI M, HUANG Y, LI C, et al. Efficient multi-stage video denoising with recurrent spatio-temporal fusion[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 3466-3475.

SONG M, ZHANG Y, AYDIN T O. Tempformer: Temporally consistent transformer for video denoising[C]//European Conference on Computer Vision. Cham: Springer Nature Switzerland, 2022: 481-496.

LI D, SHI X, ZHANG Y, et al. A simple baseline for video restoration with grouped spatial-temporal shift[C]//2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Vancouver: IEEE, 2023: 9822-9832.

LIANG J, CAO J, FAN Y, et al. Vrt: A video restoration transformer[J]. IEEE Transactions on Image Processing, 2024, 33: 2171-2182.

ZAMIR S W, ARORA A, KHAN S, et al. Multi-stage progressive image restoration[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 14821-14831.

COOLEY J W, TUKEY J W. An algorithm for the machine calculation of complex Fourier series[J]. Mathematics of Computation, 1965, 19(90): 297-301.

LECUN Y, BOSER B, DENKER J S, et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation, 1989, 1(4): 541-551.

PONT-TUSET J, PERAZZI F, CAELLES S, et al. The 2017 DAVIS challenge on video object segmentation[EB/OL]. (2018-03-01) [2024-06-20]. https://arxiv.org/abs/1704.00675v3https://arxiv.org/abs/1704.00675v3.

CLAUS M, VAN GEMERT J. Videnn: Deep blind video denoising[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops. Long Beach: IEEE, 2019: 1843-1852.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Research on a prompt learning method for intangible cultural heritage art image classification

Facial expression recognition in Swin Transformer by embedding hybrid attention mechanism

Prediction framework for drug-target interactions with multi-source descriptors fusion

Breast MRI image generation method based on generative adversarial network

Multi-label text classification combining attention with CorNet

Related Author

ZHOU Mingquan

LU Zhuoming

LIU Xinda

ZHANG Qinyu

HU Yuxia

YU Wancheng

WANG Kunxia

DING Yanrui

Related Institution

Virtual Reality Research Center of Ministry of Education, Being Normal University

College of Letters and Science, University of California

School of Information Science and Technology, Northwest University

Anhui International Joint Research Center for Ancient Architecture Intellisencing and Multi-Dimensional Modeling

School of Electronic and Information Engineering, Anhui Jianzhu University

⁰